Cilia gene mutations cause atrioventricular septal defects by multiple mechanisms

Atrioventricular septal defects (AVSDs) are a common severe form of congenital heart disease (CHD). In this study we identified deleterious non-synonymous mutations in two cilia genes, Dnah11 and Mks1, in independent N-ethyl-N-nitrosourea-induced mouse mutant lines with heritable recessive AVSDs by whole-exome sequencing. Cilia are required for left/right body axis determination and second heart field (SHF) Hedgehog (Hh) signaling, and we find that cilia mutations affect these requirements differentially. Dnah11^avc4 did not disrupt SHF Hh signaling and caused AVSDs only concurrently with heterotaxy, a left/right axis abnormality. In contrast, Mks1^avc6 disrupted SHF Hh signaling and caused AVSDs without heterotaxy. We performed unbiased whole-genome SHF transcriptional profiling and found that cilia motility genes were not expressed in the SHF whereas cilia structural and signaling genes were highly expressed. SHF cilia gene expression predicted the phenotypic concordance between AVSDs and heterotaxy in mice and humans with cilia gene mutations. A two-step model of cilia action accurately predicted the AVSD/heterotaxyu phenotypic expression pattern caused by cilia gene mutations. We speculate that cilia gene mutations contribute to both syndromic and non-syndromic AVSDs in humans and provide a model that predicts the phenotypic consequences of specific cilia gene mutations.

Introduction

Congenital heart disease (CHD), or structural heart defects present at birth, is the most common birth defect associated with significant morbidity and mortality worldwide (1–3), yet most of the genetic contribution to CHD remains undescribed (1). Atrioventricular septal defects (AVSDs [MIM: 606215]) are a class of CHD with septal defects at or adjacent to the level of the AV valves, including atrial septal defects of the primum type, ventricular septal defects of the inflow type, or complete common AV canal spanning the atria and ventricles. AVSDs comprise 5–10% of CHD and a greater proportion of cases with significant morbidity and mortality (4). Patients with AVSDs have a higher recurrence rate of CHD among their offspring compared with patients with other types of CHD (5). In familial non-syndromic CHD, the phenotypic concurrence for AVSD between affected family members is much higher than that for most other types of CHD (6,7). We previously performed a forward genetic screen in mice that demonstrated more uniform expressivity of familial AVSDs than other forms of CHD (8). These observations suggested that AVSDs might have a tractable genetic basis. We identified and propagated six murine lines with heritable recessive AVSDs (8).

The paradigm for the developmental basis of AVSDs has undergone recent revision. AVSDs were historically attributed to defects in endocardial cushion development within the heart (4). However, recent work supports a paradigm for the developmental basis of AVSDs based on defects in second heart field (SHF) cardiac progenitors (9–12). The entire atrial septum derives from the SHF and molecular events in the SHF are required for AV septation (11–15). Furthermore, morphologic analysis of AVSDs in human embryos has revealed a specific deficiency of the SHF-derived dorsal mesenchymal protrusion (DMP) (also known as the spina vestibuli or vestibular spine) (16). These observations suggest that the developmental basis of AVSDs results at least in part from deficiency of the SHF-derived DMP (17).

AV septation and DMP development require cilia-based Hedgehog (Hh) signaling (12,13), the requirement of the primary cilium for Hh signaling is well-established (18,19), and cilia are structurally present in the SHF (20). Observations consistent with a role for cilia-based Hh signaling in AV septation include: (i) SHF-specific deletion of Hh signaling genes causes AVSDs (13); (ii) The atrial septum is composed of a lineage derived from Hh-receiving SHF cardiac progenitors (12); (iii) cilia function is required for SHF Hh signaling and AV septation (18); and (iv) Hh signaling and cardiogenic transcription factors collaborate in AV septation [11). These observations suggest that cilia and Hh signaling components may contribute to the genetic basis of human AVSDs. Consistent with this hypothesis, the first human gene implicated in simplex AVSD (AVSD that present as singular defects without other organ system (syndromic) manifestations), CRELD1 (MIM: 607170) (21) has been identified as a component of cilia (http://www.sfu.ca/∼leroux/ciliome_database.htm) (22).

Cilia are also required for the establishment of embryonic L/R patterning (23). Cilia gene mutations are a cause of laterality defects, termed heterotaxy syndrome or situs ambiguous (23). Heterotaxy syndrome is characterized by randomization of the left/right body plan axis, affecting the morphogenesis of visceral organs that normally display laterality, including the heart. AVSDs and heterotaxy syndrome are strongly associated: AVSDs are the most common form of CHD in heterotaxy syndrome and approximately two-third of heterotaxy cases demonstrate AVSDs (23–28). Furthermore, a significant fraction of AVSD cases are observed in the background of heterotaxy. The degree of concurrence between heterotaxy syndrome and AVSD supports a potential common etiology (23–28). Moreover, the requirement for cilia function in both L/R determination and for AV-septum progenitor cells specification suggests that cilia function may provide a mechanistic link underlying the co-occurrence of AVSDs and heterotaxy syndrome.

In this study, we defined the genetic causation of AVSDs in two independent mouse lines, avc4 and avc6, by whole exome sequencing (WES). Line avc4 harbored a recessive mutant allele of dynein, axonemal heavy chain 11 (Dnah11) and avc6 a recessive mutant allele of Meckel syndrome, type 1 (Mks1). Dnah11^avc4 homozygous mutant mice harbored AVSDs in a minority of cases and only when situs abnormalities were observed. Dnah11 expression was absent from the wild-type embryonic SHF and heart, and Dnah11^avc4 mutant mice demonstrated normal SHF Hh signaling. In contrast, Mks1^avc6 homozygous mutant mice always showed AVSDs, although situs abnormalities were infrequent. Mks1 was strongly expressed in the murine SHF and Mks1^avc6 embryos showed a significant SHF Hh signaling decrement. The literature revealed AVSDs associated with mutations in 22 mouse and 7 human cilia genes (Tables 2 and 3). AVSD/situs concordance for Dnah11 mutants matched that reported for cilia motility gene mutations whereas AVSD/situs concordance for Mks1 mutants matched that reported for cilia structural/signaling gene mutations. We performed unbiased transcriptome analysis of the murine SHF by RNA-seq and showed that cilia motility genes were not expressed whereas the cilia structural and signaling genes were highly expressed. A model based on a requirement for cilia function in two separable roles, L/R determination and SHF Hh signaling, defines three classes of cilia gene mutations and accurately predicts the distinct expressivity of AVSDs and heterotaxy observed in mice and humans with cilia mutations.

Results

WES of mouse lines AV canal 2–6 (avc2–6)

We previously conducted a forward genetic screen and mapped N-ethyl-N-nitrosourea (ENU)-induced mutations causing AVSDs in six lines, atrioventricular canal 1–6 (avc1–6), to distinct loci on five mouse chromosomes (Table 1) (8,18). Avc1 was previously identified as a mutant allele of intraflagellar transport protein 172 (Ift172^avc1), a gene required for cilia biogenesis and Hh signaling (Table 1) (18,19). Mutation recovery in the remaining five avc lines (avc2–6) was conducted using WES analysis. We applied a filtering strategy that only retained variants that were homozygous in the affected mouse, that were absent from other strains exome-sequenced for this or other unrelated projects (15 comparator strains in total), and that were located within the mapped interval for each line, respectively. Single homozygous coding mutations were identified in avc4 and avc6 (Table 1). No homozygous coding or splice site variants were identified in lines avc3 and avc5 (Table 1). Multiple coding variants were found in the avc2 line and identification of the causative mutation is under investigation. Potential explanations for the failure to identify candidate mutations in avc3 and 5 include the possibility that the causative mutations reside in non-coding regions or unannotated coding sequence not included in the utilized exome capture design. Avc4 and 6 were prioritized for further investigations (Table 1).

Table 1.

Summary for mutations identified by WES in avc mutant lines obtained in forward genetic screen (8)

Linea	Chr	Marker 1	Marker 2	Mbpb	Nucleotide alterationsc	AA alterations	Affected gened	Predicted effect (probability)
avc1	5	D5Mit229	D5Mit420	3.6	Chr5: g.31265640A > Ge	p.Arg841Glyfs*23e	Ift172 Chr5:31253279-31291114	Mis-splicing, exon 24 skipping, a PTC in exon 25e
avc2	15	proximal end of chr5	D15MIT267	24.2	IP	IP	IP	IP
avc3	9	D9Mit106	D9Mit302	3.5	ND	ND	ND	ND
avc4	12	D12Mit181	distal end of chr12	11.5	Chr12:g.117931223T > C	p.Ser3630Pro	Dnah11 Chr12:117877982-118199043	Missense; Deleterious (PROVEAN score: -4.717; PolyPhen-2 score: 0.999)
avc5	11	D11Mit320	D11Mit245	6.1	ND	ND	ND	ND
avc6	11	D11Mit245	D11Mit360	26.5	Chr11: g.87853328 C > T	p.Arg27*	Mks1 Chr11:87853225-87863679	Non-sense; a PTC in exon 1;
								Deleterious (PROVEAN score: −3.249; PolyPhen-2 score: 0.997)

Using traditional Mendelian techniques and whole exome next generation sequencing, we have mapped mutations in four mouse lines with heritable CHD.

^amouse mutant lines generated by ENU-mutagenesis in forward genetic screen.

^bcandidate interval after mapping.

^cfor each line at least three controls (unaffected mice) and three mutants (affected mice) were sequenced, all mutations were absent in control animals.

^dchromosomal location of the affected genes was determined using the December 2011 (GRCm38/mm10) assembly from UCSC.

^emutation described in Friedland-Little et al. (18); amino acids variations with a score equal to or below −2.5 [based on PROVEAN (80)] and above 0.85 [based on PolyPhen-2 (81,82)] are predicted to be deleterious. Abbreviations: avc, atrioventricular canal; Chr, chromosome; IP, in progress; ND, none detected; PTC, premature termination codon.

Avc4 carries a mutation in a dynein, axonemal, heavy chain 11 (Dnah11) gene

Avc4 caused AVSDs (Fig. 1A and B) and was previously mapped to a 11.5-Mb region on mouse chromosome 12 between D12MIT181 and the distal end of chromosome 12 (8) (Table 1). Cardiac defects in addition to AVSD in avc4 were previously reported (8). WES of a single avc4 mutant predicted a homozygous missense mutation, T > C transition at g.117931223, within the Dnah11 gene, encoding a dynein, axonemal, heavy chain 11 (Table 1). The Dnah11 mutation was confirmed to be homozygous in E10 avc4 mutant embryos by Sanger sequencing (Fig. 1C). The T > C transition resulted in a non-conservative amino acid change, p.Ser3630Pro, located in a highly conserved ATP-binding dynein motor region D5 of the DNAH11 protein (Fig. 1D). This change was predicted to be deleterious by PROVEAN (score: −4.717) and PolyPhen-2 (score: 0.999) (Table 1). To determine whether this mutation affected transcript stability, we performed qPCR on RNA isolated from E10 avc4 wild-type and mutant embryos (three embryos/genotype). Similar levels of Dnah11 expression were observed in both genotypes (1 ± 0.652 versus 1.072 ± 0.734, P = 0.403, Student’s t test, Fig. 1E), suggesting that this mutation does not affect Dnah11 transcript stability. We concluded that the phenotype observed in avc4 mutant mice is due to Dnah11 mutation, and refer to this allele as Dnah11^avc4.

Figure 1.

Avc4 mouse line carries a recessive mutant allele of dynein heavy chain 11 (Dnah11) and Avc6 mouse line carries a recessive mutant allele of Meckel syndrome, type 1 (Mks1). (A, B) AVSDs in avc4 mutant embryos correlate with ambiguous situs. Transverse sections of E13.5 avc4 control (A) and mutant (B) hearts stained with hematoxilin and eosin. The avc4 mutant shows a large AVSD (red asterisk). (C) Representative chromatograms from one E10 avc4 wild-type (left panel C) and one E10 avc4 mutant (right panel C) from Sanger sequencing of E10 avc4 embryos (three embryos/genotype) confirmed the presence of homozygous missense mutation, T > C transition at g.117931223, within the Dnah11 gene. Red asterisk indicates a position of base substitution. (D) This single base-pair mutation results in a non-conservative amino acid change, p.Ser3630Pro (highlighted), located in a highly conserved ATP-binding dynein motor region D5 of the DNAH11 protein. (E)Dnah11 expression analysis by qPCR was performed on mRNA isolated from E10 avc4 control and mutant embryos (three embryos/genotype). Similar levels of Dnah11 expression were observed in both genotypes (mean ± SEM). (F, G) AVSDs in avc6 mutant embryos. Transverse sections of E13.5 avc6 control (F) and mutant (G) hearts stained with hematoxilin and eosin show AVSD (red asterisk) in 100% of examined mutants (28/28). No AVSDs were observed in wild-type littermates. (H) Representative chromatograms from one avc6 wild-type (left panel H) and one avc6 mutant (right panel H) from Sanger sequencing of E10 avc6 embryos (three embryos/genotype) confirmed the presence of homozygous non-sense mutation, C > T transition at g.87853328, in the Mks1 gene of avc6 mutant mice. Red asterisk indicates a position of base substitution. (I) This single base-pair mutation results in an arginine (conserved across species) being converted to a premature stop codon, p.Arg27* (highlighted). (J)Mks1 expression analysis by qPCR was performed on mRNA isolated from E10 avc6 control and mutant embryos (3 embryos/genotype). Reduced but detectable Mks1 expression was observed in avc6 mutant embryos (1 ± 0.101 versus 0.708 ± 0.099, P < 0.0001, Student’s t test). (K) Western blotting with an antibody against MKS1 on protein isolated from E10 avc6 control and mutant embryos. Avc6 mutant embryos expressed 23.69% of the amount of MKS1 compared with littermate wild-type controls after normalization to GAPDH (1.68 ± 0.109 versus 0.398 ± 0.235, P = 4.88E-05, Student’s t test). Data are mean ± SD. Magnification: (A, B and F, G) = 40×. The red arrowheads indicate presence of venous valve. Abbreviations: AVSD, atrioventricular septal defect; avc4, atrioventricular canal 4 line; avc6, atrioventricular canal 6 line.

Avc6 carries a mutation in Meckel syndrome, type 1 (Mks1) gene

Avc6 caused AVSDs (Fig. 1F and G) and was previously mapped to a 26.5-Mb region on mouse chromosome 11 flanked by D11MIT245 and D11MIT360 (8) (Table 1). Cardiac defects in addition to AVSD in avc6 were previously reported in (8). WES of a single avc6 mutant predicted a homozygous non-sense mutation, C > T transition at g.87853328, in the Mks1 gene (Table 1). Mks1 is a member of a class of B9 domain-containing proteins that is a component of the ciliary basal body (29–33). This mutation was subsequently verified as a true homozygous mutation using Sanger sequencing of E10 avc6 embryos (Fig. 1H). The C > T transition results in an arginine being converted to a premature stop codon, p.Arg27* (Fig. 1I, Table 1; PROVEAN score: −3.249, PolyPhen-2 score: 0.997), that may lead to the synthesis of a truncated protein or to non-sense-mediated mRNA decay (NMD). To determine whether this mutation affected transcript stability, a qPCR analysis was performed on RNA isolated from E10 avc6 control and mutant embryos (three embryos/genotype). Reduced but detectable Mks1 expression was observed in avc6 mutants (Fig. 1J, 1 ± 0.101 versus 0.708 ± 0.099, P < 0.0001). These data suggest that the aberrant Mks1 transcript could undergo NMD. To directly test this hypothesis, we performed western blotting with an antibody against MKS1 on protein isolated from E10 avc6 control and mutant embryos. Avc6 mutant embryos expressed 23.69% of the amount of MKS1 compared with littermate wild-type controls, normalization to GAPDH (1.68 ± 0.109 versus 0.398 ± 0.235, P = 4.88E-05; Fig. 1K). We speculate that the small residual expression of wild-type-size MKS1 detected in avc6 mutants may be due to translational read-through. We concluded that the phenotype observed in avc6 mutant mice is due to Mks1 mutation, and refer to this allele as Mks1^avc6.

Mks1^avc6 but not Dnah11^avc4 embryos have SHF Hh signaling defects

We investigated the mechanism underlying AVSDs in Dnah11^avc4 and Mks1^avc6 embryos. We asked whether either mutation abrogated SHF cilia-based Hh signaling, known to be required for AV septation (11–15). We observed that Dnah11 expression was undetectable in the SHF or the heart tube during this process, at E10, by whole-mount in situ hybridization in wild-type embryos (Fig. 2A and B). Dnah11 expression was observed in the forebrain, midbrain and hindbrain; regions known to have motile cilia (Fig. 2A and B). Although this result suggested that Dnah11 was not required for the SHF Hh signaling, we directly evaluated Hh signaling in Dnah11^avc4 mutants.

Figure 2.

Dnah11^avc4 embryos do not have SHF Hh signaling defects. (A, B) Expression of Dnah11 analysed in E10 wild-type embryos by whole-mount in situ hybridization. Dnah11 expression was undetectable in the SHF (blue arrowhead) and the heart tube (red arrowhead) in E10 wild-type embryos (A, B) but it was observed in the forebrain, midbrain, and hindbrain, regions known to have motile cilia (A, B). Expression of Gli1 analysed in E10 Dnah11^avc4 control (C, E) and mutant (D, F) embryos by whole-mount in situ hybridization. Similar levels of Gli1 expression were observed in both genotypes. The blue arrowheads indicate the SHF and the red arrowheads—the heart. (G)Gli1 and Ptch1 expression analysis by qPCR was performed on mRNA isolated from SHF microdissected from E10 Dnah11^avc4 wild-type and mutant embryos. Similar levels of Gli1 and Patch1 expression were observed in both genotypes (1 ± 0.10 versus. 0.98 ± 0.16, P = 0.337 for Gil1; 1 ± 0.11 versus 1.032 ± 0.20, P = 0.279 for Ptch1, Student’s t test). Full-length (Gli3–190) and processed (Gli3–83) forms of Gli3 were detected by western blot in Dnah11^avc4 wild-type and mutant embryos (H,I). The Gli3-83/Gli3-190 ratio was statistically indistinguishable between individual Dnah11^avc4 wild-type and mutant embryos, with similar degrees of variability observed in both sample sets (1 ± 0.266 versus 1.04 ± 0.240, for wilt-type vs. Dnah11^avc4, respectively, P = 0.874, Student’s t test) (H, I). Data are mean ± SD. Magnification: (A–F) = 40×. Abbreviations: avc4, atrioventricular canal 4 line; SHF, second heart field.

We quantitatively assessed the integrity of the Hh signaling pathway in Dnah11^avc4 mutants by expression of Hh pathway target genes and biochemically. We first evaluated the Hh-dependent expression pattern of the Hh pathway genes Gli1 and Patched1 (Ptch1). Expression levels of Gli1 and Ptch1 reflect Hh signaling activity (18,34,35). SHF Gli1 expression appeared qualitatively similar in wild-type and Dnah11^avc4 mutant embryos at E10 by whole-mount in situ hybridization (Fig. 2C–F). Gli1 and Ptch1 expression in the posterior SHF (pSHF), microdissected at E10, were assessed quantitatively by qPCR. We found that Gli1 and Ptch1 transcription levels in the pSHF were statistically the same in wild-type and Dnah11^avc4 embryos (1 ± 0.10 versus 0.98 ± 0.16, P = 0.337 for Gil1; 1 ± 0.11 versus 1.032 ± 0.20, P = 0.279 for Ptch1; Fig. 2G). Because the qPCR gene expression analysis required samples pooled from three distinct embryonic pSHFs, it remained feasible that individual sample variability in Hh signaling, that could explain the variable expression of AVSDs in Dnah11^avc4 embryos, could have been masked. Therefore, we quantitatively assessed Hh signaling in individual thoracic samples by examining the proteolytic processing of the Gli3 transcription factor. In mammals, Gli3 exists in two forms, a full-length transcriptional activator (Gli3-190), and a proteolytically processed transcriptional repressor (Gli3-83). Processing of Gli3 from the full-length to the truncated form is regulated by Hh signaling and the ratio of Gli3-83/Gli3-190 is a quantitative measure of Hh signaling activity (18). We found that the Gli3-83/Gli3-190 ratio was statistically indistinguishable between individual wild-type and Dnah11^avc4 embryos, with similar degrees of variability observed in both sample sets (1 ± 0.266 vs. 1.04 ± 0.240, for wild-type vs. Dnah11^avc4, respectively; Fig. 2H and I). No single Dnah11^avc4 mutant embryo displayed a quantitative Hh signaling decrement that could cause an AVSD (18). These results supported the conclusion that Dnah11^avc4 does not cause a decrement of SHF Hh signaling, and that Dnah11^avc4 must affect AV septation by another mechanism.

In contrast, we observed a significant SHF Hh signaling defect in Mks1^avc6 embryos. Strong SHF Mks1 expression was observed at E10 by whole-mount in situ hybridization (Fig. 3A and B). We observed a severe decrement of the expression of Hh signaling target Gli1 in the SHF of Mks1^avc6 mutant embryos by whole-mount in situ hybridization at E10 (Fig. 3D and F versus C and E). This SHF Hh signaling defect provided a mechanistic explanation of the penetrant AVSDs observed in the Mks1^avc6 line. Mks1 expression was excluded from the developing heart in wild-type embryos (Fig. 3A and B), consistent with the required role for Hh signaling in SHF cardiac progenitors, not the heart, for AV septation (12).

Figure 3.

Mks1^avc6 embryos have SHF Hh signaling defects. (A, B) Expression of Mks1 analysed in E10 avc6 wild-type embryos by whole-mount in situ hybridization. Mks1 was expressed in the SHF (blue arrowhead) but not in the heart tube (red arrowhead) in E10 avc6 wild-type embryos (A, B). Expression of Gli1 analysed in E10 Mks1^avc6 control (C, E) and mutant (D, F) embryos by whole-mount in situ hybridization. A severe qualitative decrement of Gli1 expression was observed in the SHF of Mks1^avc6 mutant embryos. The blue arrowheads indicate the SHF and the red arrowheads – the heart. Magnification: 40×. Abbreviations: avc6, atrioventricular canal 6 line; SHF, second heart field.

Dnah11^avc4 mutants demonstrated AVSDs only with disrupted laterality

Since Dnah11^avc4 embryos demonstrated normal SHF Hh signaling, another deficit must have caused the observed AVSDs. We hypothesized that Dnah11^avc4 caused AVSDs by cardiac isomerism resulting from laterality defects. Dnah11 encodes a member of the cilia motility apparatus (36–38) and targeted disruption of the mouse Dnah11 (lrd) gene results in randomization of laterality (37,39,40). DNAH11 mutations in humans cause primary cilia dyskinesia (PCD) and Kartagener syndrome (36,41–44). The Dnah11^avc4 mutation is distinct from the previously reported Dnah11 mutations in both, mice (37,40,45) (Table 2) and humans (37,41,43,44) (Table 3). We investigated the relationship between Dnah11^avc4, AV septation, and body plan situs. 58% of avc4 mutants (29/50; Supplementary Material, Table S1) demonstrated situs solitus and normal AV septation was observed in each situs solitus mutant (Supplementary Material, Table S1). Situs inversus totalis was observed in 24% of avc4 mutant embryos (12/50; Supplementary Material, Table S1 and data not shown) and normal AV septation was observed in each situs inversus totalis mutant (Supplementary Material, Table S1). 18% of avc4 mutants (9/50; Supplementary Material, Table S1 and data not shown) demonstrated abnormal situs with dextrocardia (4/9) and/or lung isomerism (6/9). AVSDs were present in all of these nine mutant embryos (Fig. 1B versus A; Supplementary Material, Table S1). Additionally, serial cardiac section analysis suggested atrial isomerism, with (i) venous valves that are a normal feature of the right atrium observed in both atria of mutant mice with AVSDs (right atrial isomerism, 3/9) (Fig. 1B versus A) or (ii) absence of venous valves in both atria of mutant mice with AVSDs (left atrial isomerism, 5/9, data not shown). Littermate wild-type control animals all demonstrated situs solitus and normal AV septation (14/14; Fig. 1A and data not shown). AVSDs in avc4 mutant embryos were significantly associated with abnormal or ambiguous situs (Fisher’s Exact Test (FET), P < 0.0001, Supplementary Material, Table S1).

Table 2.

List of known mouse cilia genes mutations causing AVSD

Gene	Alleles	Location and MGI	AVSD/situs correlation	References
Dnaic1	Dnaic1tm1.2Leo (c); Dnaic1b2b1526Clo; Dnaic1b2b284Clo (a)	4A5, MGI:1916172	AVSD with heterotaxy only	(24,62,63,88–94)
Dnah5	Dnah5b2b002Clo; Dnah5b2b016Clo; Dnah5b2b601Clo; Dnah5b2b1154Clo; Dnah5b2b2395Clo; Dnah5b2b1003Clo; Dnah5hlb612(a);	15B1, MGI:107718	AVSD with heterotaxy only	(25,56,59,62,63,88,90–93,95–100)
Dnah11	Dnah11b2b1279Clo; Dnah11b2b1727Clo; Dnah11b2b2349Clo; Dnah11tm1Ssp (a); Dnahc11lrm3(a); Dnah11iv (a)	12F2, MGI:1100864	AVSD with heterotaxy only	(27,28,37,39,40,45,56,62,63,88,90,93, 96,101–107)
Ccdc39	Ccdc39b2b1735Clo; Ccdc39b2b2025.1Clo	3A3, MGI:1289263	AVSD with heterotaxy only	(62,63,88)
Drc1	Drc1b2b1654Clo	5B1, MGI:2685906	AVSD with heterotaxy only	(62,63,88)
Hspb11/Ift25	Hspb11tm1b(EUCOMM)Wti (c); Hspb11tm1a(EUCOMM)Wtsi,(c);	4C7, MGI:1920188	AVSD without heterotaxy	(108,109)
Ift88	Ift88cbbs (b);	14C, MGI:98715	AVSD without heterotaxy	(110,111)
Ift172	Ift172avc1(b)	5B1, MGI:2682064	AVSD without heterotaxy	(8,18,112)
Sufu	Sufub2b273Clo,	19C3, MGI:1345643	AVSD without heterotaxy	(61–63,66,88)
Kif7	Kif7b2b2254Clo; Kif7maki; Kif7tm1.1Hui (c)	7D2, MGI:1098239	AVSD without heterotaxy	(61–64,88,112)
Fuz	Fuzb2b1273Clo; Fuzgt/gt (b)	7B2, MGI:1917550	AVSD without heterotaxy,	(62,63,65,88,112–114)
Cntrl/Cep110	Cntrlb2b1468Clo	2B-C1, MGI:1889576	AVSD without heterotaxy	(62,63,88)
Mks1	Mks1avc6(b);	11C11, MGI:3584243	AVSD without heterotaxy	(8) Fig. 1F and G, Supplementary Material, Table S1
Dync2h1	Dync2h1b2b414Clo; Dync2h1Gt(RRM278)Byg; Dync2h1lln	9A1, MGI:107736	AVSD irrespective of heterotaxy	(62,63,88,112,115–117)
Mks1	Mks1hlb614; Mks1tm1a(EUCOMM)Wtsi (c);	11C11, MGI:3584243	AVSD irrespective of heterotaxy	(29,31,32,69,72,73,98,118–124)
B9d1	B9d1tm1a(EUCOMM)Wtsi (c); B9d1tm1d(EUCOMM)Wtsi (c)	11B2, MGI:1351471	AVSD irrespective of heterotaxy	(98,124–126)
Cc2d2a	Cc2d2ab2b1035Clo	5B3, MGI:1924487	AVSD irrespective of heterotaxy	(53,62,63,88,98)
Nphp3	Nphp3tm1Cbe (c)	9F1, MGI:1921275	AVSD irrespective of heterotaxy	(127)
Pskh1	Pskh1b2b1230Clo	8D3, MGI:3528383	AVSD irrespective of heterotaxy	(62,63,88)
Tbc1d32	Tbc1d32b2b2284Clo; Tbc1d32b2b2596Clo; Tbc1d32bromi (a)	10B3, MIM:2442827	AVSD irrespective of heterotaxy	(62,63,88,128)
Ift27	Ift27tm1b(EUCOMM)Hmgu (c)	15E2, MGI:1914292	AVSD irrespective of heterotaxy	108, 109
Ift74	Ift74b2b796Clo	4C5, MGI:1914944	AVSD irrespective of heterotaxy	(62,63,88,129,130)
Ift88	Ift88fxo (b); Ift88tm1.1Bky (c); Polaristm1Rpw (c); Ift88tm1Rpw (c)	14C, MGI:98715	AVSD irrespective of heterotaxy	(19,56,57,110–113,115,116,131–142)
Ift172	Ift172wim (a)	5B1, MGI:2682064	AVSD irrespective of heterotaxy	(18,19,64,112,115–117)
Ift140	Ift140b2b1283Clo	17A3.3, MGI:12146906	AVSD irrespective of heterotaxy	(62,63,88)

Alleles type: ^(a), null; ^(b), hypomorph; ^(c), targeted (null/knockout). Abbreviation: AVSD, atrioventricular septal defect.

Table 3.

List of known human cilia genes mutations causing AVSD

Gene	Other aliases	Location and OMIM	Human syndrome	AVSD/situs correlation	References
DNAI1	CILD1, DIC1, ICS1,	9p13.3, OMIM:604366	PCD, Kartagener syndrome	AVSD with heterotaxy only	(23,42,59,90–93,95–97, 143–146)
DNAH5	CILD3, DNAHC5	5p15.2, OMIM:603335	PCD, Kartagener syndrome	AVSD with heterotaxy only	(23,42,59,90–97,143, 147–151)
DNAH11	CILD7, DNAHC11	7p21, OMIM:603339	PCD, Kartagener syndrome	AVSD with heterotaxy only	(36,37,41–44,59,90–93, 95–97)
MKS1	MES, MKS, BBS13,	17q22, OMIM:609883	Meckel-Gruber syndrome, type 1, (MKS1); Bardet-Biedl syndrome, type 13 (BBS13)	AVSD irrespective of heterotaxy	(32,33,46,57,58,67,69, 71–73,152–154)
MKKS	KMS, MKS, BBS6	20p12.2, OMIM:604896	Bardet-Biedl syndrome McKusick-Kaufman syndrome	AVSD irrespective of heterotaxy	(57,58,67,68,155–167)
EVC	DWF-11, EVCL	4p16.2 OMIM:604831	Ellis–van Creveld syndrome	AVSD irrespective of heterotaxy	(55,57,58,164,168–179)
EVC2	LBN	4p16.1 OMIM:607261

Abbreviation: AVSD, atrioventricular septal defect; PCD, primary ciliary dyskinesia.

Mks1^avc6 embryos demonstrated AVSDs independent of laterality defects

Mks1 encodes a centrosomal protein required for primary cilium formation (30–33) and has been previously implicated in left-right patterning of mouse and human embryos (29,33) (Tables 2 and 3). The mutation found in avc6 mouse line has not been previously reported among Mks1 mutations in mice (29,31) (Table 2) or humans (30,33,46–48) (Table 3). We investigated the relationship between Mks1^avc6, AV septation, and body situs. 89% of avc6 mutant embryos demonstrated situs solitus (25/28; Supplementary Material, Table S1 and data not shown), whereas 11% of avc6 mutants showed abnormal situs with dextrocardia (3/28; Supplementary Material, Table S1 and data not shown). AVSDs were present in 100% of mutant embryos (28/28; Supplementary Material, Table S1, Fig. 1G versus F, and data not shown). All avc6 wild-type control littermate embryos showed situs solitus with normal AV septation (12/12, Fig. 1F, and data not shown). These results demonstrated that Mks1^avc6 causes AVSDs independent of L/R defects (FET, P > 0.05, Supplementary Material, Table S1) since there was no observed association between heterotaxy and AVSDs in this line (FET, P > 0.05, Supplementary Material, Table S1).

Cilia structural genes and signaling genes, but not cilia motility genes are expressed in the SHF

Our analysis of avc4, avc6 and previously avc1, suggested a correspondence between the class of cilia mutation, SHF gene expression and SHF Hh signaling. Specifically, Dnah11, a cilia motility gene, was not expressed in the SHF nor required for SHF Hh signaling whereas Mks1 and Ift172, primary cilia structural genes, were both expressed in the SHF and required for SHF Hh signaling. We asked whether this pattern of cilia gene expression was indicative of a general pattern such that the subcategory of cilia gene action predicted its SHF gene expression and requirement for SHF Hh signaling. To interrogate SHF expression of cilia genes in an unbiased manner, we performed RNA-seq on the embryonic pSHF, containing AV septum progenitors. We isolated the pSHF by microdissection from mixed background wild-type embryos at E9.5 (49). Sufficient RNA was collected by pooling the pSHF from three embryos per sample and RNA-seq was performed in quintuplicate, with each library sequenced to a depth of 17–26 million reads.

We found that genes encoding components of the cilia motility machinery were expressed at low or absent levels whereas genes encoding components of the primary cilia or cilia signaling were expressed at high levels in the SHF (Fig. 4, Supplementary Material, Table S2). Specifically, we examined the expression of the 13 axonemal dynein orthologues essential for cilia motility and whose mutations disturb L/R axis formation. We found that 12/13 cilia motility genes were negligibly expressed in the SHF (range: undetectable to 41 ± 16) consistent with transcriptional noise or absent expression (Fig. 4A and B, Supplementary Material, Table S2), utilizing an expression threshold composed of genes previously examined by in-situ hybridization and RT-PCR and defined as not expressed in the pSHF (e.g. Hoxb8, Hoxc8, Hoxd8 (50) at 41 ± 9, 5 ± 3 and 8 ± 5, respectively). In contrast, we examined the expression of cilia genes encoding intraflageller transport proteins (IFTs), required for primary cilia structure and function essential for cilia-based signaling. We found that 15/15 IFT genes identified in the ciliome database (http://www.sfu.ca/∼leroux/ciliome_database.htm) (51) and detected in our RNA-seq were highly expressed in the SHF (range: 87 ± 27 to 857 ± 93) (Fig. 4A and B, Supplementary Material, Table S2). The high levels of transcript reads of the IFT genes were comparable to genes previously examined and defined as highly expressed in the pSHF (e.g. Osr1 (11), and Foxf2 (49) at 1012.7 and 284.5, respectively). These observations demonstrated that the class of cilia genes specific to cilia motility, whose mutations disturb L/R axis formation, are not expressed in the SHF and are therefore unlikely to be required for SHF Hh signaling, whereas all primary cilia genes are strongly expressed in the SHF and may facilitate SHF Hh signaling.

Figure 4.

Cilia structural genes and cilia signaling genes, but not cilia motility genes, are expressed in the SHF. (A) All genes detected by RNA-seq plotted by log (mean expression), with genes from Tables 2 and 3 labeled by phenotypic classes (orange, AVSD with heterotaxy only; blue, AVSD without heterotaxy; green, AVSD irrespective of heterotaxy). Genes below a normalized mean expression n of 50 fall within low-level transcriptional noise. (B) Graph of Dnah and Ift cilia genes expressed by respective normalized FPKM. (C) Graph of all cilia genes expression by phenotypic class with respective normalized FPKM. (D) qPCR expression analysis of 22 cilia genes associated with AVSD in mice and humans (Tables 2 and 3) was performed on mRNA isolated from E9.5 wild-type whole mouse embryo and pSHF. Low expression of cilia genes whose mutations caused AVSD with heterotaxy only was observed in contrast to high expression of all of the genes whose mutations caused AVSD without heterotaxy or AVSD irrespective of heterotaxy. qPCR data are mean ± SD, n = 3, *P < 0.01, **P < 0.001, Student’s t test.

AVSDs are a common feature in mice and humans with cilia mutations

Our analysis of avc1 (8,18), avc4 and avc6 indicated a strong relationship between AVSD causation and cilia gene mutations. We reviewed the mouse and human cilia literature to determine whether there was an association between cilia gene mutations and AVSDs, including primum ASDs, inlet VSDs, common atrium, and complete common AV canal. Review of the mouse cilia literature showed that cilia mutations commonly caused AVSDs. Specifically, we identified 22 cilia genes with alleles that caused AVSDs as part of the phenotypic spectrum (Table 2). Human ciliopathies are a group of over 100 overlapping clinical disorders caused by defects in the cilium and its anchoring structure, the basal body (52–57). We found that specific human ciliopathy mutations, including those causing Ellis–van Creveld, McKusick-Kaufman, Bardet-Biedl type 13, Meckel-Gruber, and Kartagener syndrome (58), were commonly associated with AVSDs (Table 3).

AVSDs/heterotaxy correspondence in humans and mice with cilia gene mutations is predicted by SHF gene expression

We hypothesized that AVSD/heterotaxy concordance caused by cilia gene mutations would be predicted by the cilia gene’s SHF gene expression. We compared the SHF RNA-seq expression values of the 26 cilia genes associated with AVSDs in mice and humans (Tables 2 and 3) with the AVSD/heterotaxy phenotypic concordance caused by their mutations (Fig. 4A and C, Supplementary Material, Table S1), and validated the relative pSHF expression for 22 by qPCR (Fig. 4D). AVSDs caused by mouse or human mutations in cilia genes with roles selective to cilia motility (e.g. Dnaic1, Dnah5 or Dnah11 in mice and DNAH5 [MIM: 603335] or DNAH11 [MIM: 603339] in humans) were presented only concurrently with heterotaxy syndrome (Tables 2 and 3). We found that the SHF expression for these cilia genes was low or undetectable (Fig. 4). In contrast, AVSDs caused by cilia mutations affecting the Hh signaling were observed without heterotaxy or irrespective of heterotaxy (Table 2). Cilia genes whose mutations caused AVSD without or irrespective of heterotaxy were strongly expressed in the SHF. Expression of cilia genes whose mutations caused AVSDs only with heterotaxy was significantly lower than that for cilia genes whose mutations caused AVSD either without heterotaxy or irrespective of heterotaxy (Kruskal-Wallis H stat = 12.16, P = 0.0023; AVSD with heterotaxy only versus AVSD without heterotaxy (P = 0.0336) and AVSD with heterotaxy only versus AVSD irrespective of heterotaxy (P = 0.0016)) (Fig. 4A and C). These results demonstrated perfect concordance between SHF gene expression and the pattern of AVSDs and heterotaxy syndrome presentation in mice and humans with cilia mutations.

Discussion

We report the genetic basis of AVSDs in two ENU-induced mutant mouse lines, avc4 and avc6. A Dnah11 mutation segregated with AVSDs in line avc4, and a Mks1 mutation segregated with AVSDs in line avc6. Because these mutations were the only homozygous mutations identified in embryos demonstrating recessive AVSDs, we conclude that Dnah11^avc4 is causative of AVSDs in line avc4 and that Mks1^avc6 is causative of AVSDs in line avc6. These cilia gene mutations join the identification of Ift172 mutation in avc1 (18), defining cilia gene mutations as causative of AVSDs in three murine avc lines analysed to date. Cilia gene mutations are not a well-appreciated cause of AVSDs, perhaps because AVSD presentation is part of a complex group of phenotypes often observed in ciliopathies. Nonetheless, our review of the literature identified mutations in 22 mouse cilia genes and 7 human cilia genes with AVSDs described as part of the phenotypic spectrum. These observations suggest that although not responsible for all AVSDs, mutations in cilia genes may play an important role in AVSD causation.

Primary and motile cilia are both required for L/R determination (54,55,57). Furthermore, our findings suggest that primary cilia but not motile cilia are required for SHF Hh signaling (Figs. 2 and 3). Based on these two distinct cilia-based processes, we generated a model that predicted three classes of cilia gene mutations causing distinct expression of AVSDs and heterotaxy (Fig. 5): (i) Mutations that affect laterality but not SHF Hh signaling; (ii) mutations that affect SHF Hh signaling but not laterality; and (iii) mutations that affect both laterality and SHF Hh signaling. We found that the phenotypic expression of mutant mouse alleles from our forward genetic screen and more generally from mouse and human cilia alleles described in the literature (Tables 2 and 3) support this model, as detailed below.

Figure 5.

Three classes of cilia mutations with distinct effects on laterality and Hh signaling cause distinct presentations of AVSDs and heterotaxy. Class 1 mutations affect cilia motility (resulting in disruption of correct L/R axis formation) but not AV septum progenitors. Mutations in this class, such as Dnah11^avc4, only cause AVSDs secondary to abnormal situs, when present. Class 2 mutations affect signaling but not laterality. Mutations of this class, such as Ift172^avc1 or Mks^avc6, disrupt AV septum progenitors causing AVSDs but not situs abnormalities. Class 3 mutations disrupt both cilia motility and signaling. Mutations of this class, such as Polaris, disrupt laterality and AV septum progenitors and cause AVSDs irrespective of situs abnormalities. This three-class model accurately fits the phenotypic variance of AVSDs and heterotaxy observed in humans and mice with cilia gene mutations.

The model (Fig. 5) predicts that mutations of cilia genes with roles specific to cilia motility would present AVSDs only concurrently with heterotaxy. In this report, we found that AVSDs in Dnah11^avc4 mutant mice occurred only with heterotaxy (Figs 1A, B and 5, Table 2, Supplementary Material, Table S1). Mutations of Dnah11 or other genes specifically required for cilia motility, such as Dnaic1 and Dnah5, cause randomized L/R patterning (24,25,27,28,37,39,40). Mutations in their human orthologues cause Primary Ciliary Dyskinesia (PCD), a disorder also characterized by decreased or absent cilia movement, including respiratory morbidity and impaired male fertility (23,36,37,42–44,59). PCD patients can present with normal or inverted laterality (e.g. Kartagener Syndrome) or heterotaxy syndrome (37,41–44,59). AVSDs were only observed in PCD patients with heterotaxy syndrome, not in humans with PCD or in mice with mutations in cilia motility genes when normal or perfectly inversed situs was present (36,37,42–44,59), (Tables 2 and 3, Supplementary Material, Table S1) analogous to the AVSDs observed in the Dnah11^avc4 line (Fig. 1B, Table 2, Supplementary Material, Table S1). These observations imply that cilia motility genes are not required for SHF Hh signaling, otherwise mutations in this class of cilia gene would cause penetrant AVSDs regardless of their impact on situs. In this report, we found that Dnah11 is not expressed in the SHF (Fig. 2A and B) and Dnah11^avc4 does not disrupt SHF Hh signaling (Fig. 2C–I). Remarkably, we also observed that the entire cohort of dynein heavy chains and other genes specifically required for cilia motility is negligibly expressed in the SHF (Fig. 4, Supplementary Material, Table S2). These observations provide direct evidence that the cilia motility machinery is not required for SHF Hh signaling.

The perfect correspondence between AVSDs and L/R abnormalities caused by mutations of cilia genes required specifically for cilia motility implies that AVSDs result from situs abnormalities in these cases. The atrial septum is derived entirely from the left side of the wild-type SHF, demonstrated by its inclusion in the fate map of the left-sided marker Pitx2 (60). It is therefore not surprising that mutations causing heterotaxy or sidedness defects concomitantly affect the development of this ‘sided’ structure (Supplementary Material, Table S1 and reference (27)). The molecular basis of the AVSDs caused by heterotaxy awaits future investigation, although this study appears to rule out a role for SHF Hh signaling defects.

The model (Fig. 5) predicts that mutations in cilia genes with a selective role in Hh signaling but not laterality would be predicted to cause AVSDs without heterotaxy (Fig. 5). This cohort includes reported mutations in Sufu and Kif7 (61–66) (Tables 2 and 3).

The model (Fig. 5) predicts that mutations in cilia genes that affect Hh signaling and L/R determination would cause AVSDs irrespective of heterotaxy (Fig. 5). This group includes mutations in multiple intraflagellar transport genes, such as Ift88^tm1Rpw (Polaris) and other genes that affect primary cilia function. This class of mutations causes defective Hh signaling, resulting in AVSDs in all cases, and defective L/R determination, resulting in randomized laterality and heterotaxy in some cases. The independent effect of these cilia gene mutations on SHF Hh signaling and L/R determination uncouples the phenotypic expression of AVSDs and heterotaxy.

Interestingly, a select group of mutations in genes with roles in both Hh signaling and laterality also cause AVSDs without affecting laterality. For example, although Mks1 is required for ciliogenesis and Hh signaling, and severe Mks1 mutations cause both laterality and Hh signaling defects in mice (29,31), we observed that the Mks1^avc6 allele significantly caused AVSDs (28/28) but not heterotaxy (3/28; P > 0.05) (Figs. 1F, G and 5, Supplementary Material, Table S1). Mks1 is expressed in the SHF at E10 (Fig. 3A and B) and is required for SHF Hh signaling (Fig. 3C–F). These observations suggest that the avc6 allele of Mks1 is hypomorphic, perhaps explained by the small amount of residual wild-type-sized MKS1 observed in avc6 mutant embryos (Fig. 1K). Thus, Mks1^avc6 affects SHF Hh signaling with high penetrance but L/R axis determination with much lower penetrance. In humans, the literature demonstrates an allelic series of Mks1 mutations; more severe Mks1 mutations cause both AVSDs and situs abnormalities whereas milder mutations cause AVSDs in the absence of situs abnormalities (29,31–33,46–48,67–73). We also observed this pattern among Ift172 alleles in mice: Ift172^avc1, a hypomorphic allele, caused Hh signaling but not situs defects (18) whereas Ift172^wim, a protein-null, caused Hh signaling and situs abnormalities (19). These observations suggest that for the genes that contribute to primary cilia function, the requirement for Hh signaling may be more dosage-sensitive than the requirement for L/R determination. The molecular basis for this distinction is currently unclear. It is possible that cilia in the node and the SHF may have distinct structural or signaling properties. Based on these examples, we speculate that hypomorphic mutations of the large number of genes that contribute to primary cilia structure and signaling may contribute to human simplex AVSDs.

The relationship between cilia gene mutations and the genetic underpinnings of human simplex AVSDs, the most common form of AVSDs in CHD, awaits future studies. A-priori, our model predicts that mutations affecting cilia-based Hh-signaling, but not laterality, likely predominate. The first single-gene risk factor for simplex AVSD in humans, CRELD1 (74), encodes a cilia-localized protein. Consistent with our model, CRELD1 does not appear to be required for L/R determination, since Creld1 null mice do not manifest L/R abnormalities (75), suggesting that CRELD1 mutations contribute to the cause of AVSD through disruption of SHF Hh signaling rather than cilia motility. Our findings coalesce with previous literature to provide a conceptual scaffold for understanding the pleiotropic effects of cilia gene mutations in human CHD. We speculate that cilia gene mutations will contribute to simplex AVSDs in humans. We also suggest that AVSDs should be considered as a part of the ciliopathy spectrum and that the clinical sequela of impaired cilia function be considered in the evaluation of patients with AVSDs.

Materials and Methods

Mouse strains

Avc1–Avc6 mouse lines were identified in a screen for recessive ENU-induced mutations that caused prenatal lethality and structural heart defects (8). All experiments employed age-, gender- and genetic strain-matched controls to account for any variations in data sets compares across experiments. All experiments were performed under the University of Chicago Institutional Animal Care and Use Committee (IACUC) approved protocol (ACUP no. 71737) and in compliance with the USA Public Health Service Policy on Humane Care and Use of Laboratory Animals.

WES analysis

Exome captures were performed on mouse genomic DNA using a hybrid capture reagent from Roche Nimblegen (SeqCap EZ Mouse Exome SR; 54.3 Mb target including 203 225 exonic regions (C57BL/6J, NCBI37/mm9); described in Fairfield et al. (76). Library construction, hybrid capture and sequencing were performed as described in Fairfield et al. (76) and Buchovecky et al. (77), with sample indexes incorporated prior to capture. All sequencing of post-enrichment shotgun libraries was performed on an Illumina GA2x, with paired-end 76-p reads (PE76), and a third read to determine the index sequence associated with each read-pair. Sequence reads were mapped to the mm9 reference genome with BWA (78), and then separated by sample based on the index sequence. Variants were called with SAMtools (79). Custom scripts were used to annotate variants with respect to their predicted impact on protein sequence. To remove inbred strain polymorphisms as well as systematic sequencing artifacts, we removed from consideration variants identified in any of the other strains described here or 10 other mice representing unrelated phenotypes. Exome sequencing identified 2973 exonic variants for avc2, 25 920 for avc3, 3532 for avc4, 2400 for avc5 and 24 331 for avc6 mutants. Strain-specific homozygous, protein-altering (missense, non-sense, canonical splice and coding indel) variants within the mapping interval were identified and subjected to Sanger confirmation.

Sanger validation of candidate mutations

Candidate mutations were validated by PCR amplification and sequencing of affected and unaffected samples from the mutant colonies (76). Briefly, genomic DNA (6 ng) was amplified in a 50 µl PCR reaction with 100 nM of specifically designed primers near the mutation site and 25 μl of Taq 2× master mix (New England Biolabs) at 94 °C for 3 min, followed by 35 cycles of 94 °C for 30 s, 57 °C for 30 s and 72 °C for 30 s, and final extension at 72 °C for 5 min. The PCR products were then purified with Qiaquick columns (Qiagen), and sent for sequencing to the University of Chicago Sequencing Facility. Primer sequences used in this study are listed in Supplementary Material, Table S3.

Interpretation of novel missense single nucleotide variations (SNVs)

To predict whether the candidate SNVs would have deleterious effects or not, we used 2 software programs, i.e., Protein Variation Effect Analyzer (PROVEAN; J. Craig Venter Institute, San Diego, California, USA, http://provean.jcvi.org/index.php) (80) and Polymorphism Phenotyping v2 (PolyPhen-2; Harvard University, Cambridge, Massachusetts, US, http://genetics.bwh.harvard.edu/pph2) (81,82). PROVEAN uses sequence homology to predict amino acid substitutions that will affect protein function, thus contributing to a disease (80). PROVEAN predicts substitutions with a score ≤ −2.5 as being ‘deleterious’, and with a score above the threshold −2.5 as being ‘neutral’. PolyPhen-2 takes into account the physicochemical characteristics of the wild-type and mutated amino acid residue and the consequence of the amino acid change for the structural properties of the protein in addition to evolutional conservation (81,82). PolyPhen-2 generates a different scale of reported scores, with the corresponding predictions being ‘probably damaging’ with a score larger than 0.85, ‘possibly damaging’ with a score between 0.85 and 0.15, and ‘benign’ with a score <0.15. Because PolyPhen-2 considers only human protein sequences, the mouse SNVs were investigated in the context of human protein sequences.

Histology

Embryos harvested at E13.5 were flushed with cold PBS, genotyped and fixed overnight at 4 °C in 4% paraformaldehyde (Sigma-Aldrich). Subsequently, they were processed for paraffin-embedded sections and analysed by H&E staining according to the manufacturer’s protocol (Sigma-Aldrich).

RNA isolation and RT-PCR

To examine the effect of mutations on transcripts, we extracted total RNA from control and mutant E10 embryos using RNeasy Mini Kit (Qiagen) according to the manufacturer’s protocol. Reverse transcription reaction was performed using the SuperScript III First-Strand Synthesis SuperMix for quantitative RT-PCR (Invitrogen) according to the manufacturer's recommendations. RT-PCR was performed using the POWER SYBR Green PCR master mix from Applied Biosystems and run on an Applied Biosystems AB7500 machine in 96 well plates. The relative gene expression level was calculated by the ΔΔCt method (83) using glyceraldehyde-3-phosphate dehydrogenase (Gapdh) gene expression level as internal control. The data presented are the average of three independent experiments. Primer sequences used in this study are listed in Supplementary Material, Table S3.

In situ hybridization

Whole-mount RNA in situ hybridization was performed as described previously (84). Briefly, sense and antisense probes were generated using a digoxigenin (DIG) RNA labeling kit (Roche). Probes were hybridized overnight at 65 °C with E10 wild-type embryos. DIG-labeled probes were detected by anti-digoxin-AP Fab fragments (Roche) and precipitated by BM purple AP substrate (Roche). Primer sequences used to generate Dnah11 and Mks1 ISH probes are listed in Supplementary Materials, Table S3. The Gli1 ISH probe was a kind gift from Dr Elizabath Grove (University of Chicago, Chicago, IL, USA).

Western blotting

Western blots were performed in sextuplicate, Detection of GLI3 and MKS1 protein was performed according to standard procedures (18) using a polyclonal rabbit anti-GLI3 antibody (sc-20688; Santa Cruz Biotechnology, Santa Cruz, CA) and polyclonal rabbit anti-MKS1 antibody (16206-1-AP; Proteintech Group Inc., Chicago, IL).

RNA-seq data analysis

Total RNA was extracted from the pSHF as described previously in (49), and five libraries were generated by pooling 3 samples per library (TruSeq RNA Sample prep kit v2; Part no. RS-122-2001). 51-bp single-ended sequencing libraries were prepared and sequenced using the Illumina HiSeq2500 platform by the Genomics Core Facility at the University of Chicago (Invitrogen, 2013). We focused on the 38-bp reads on the right-side with a general declining pattern of quality scores as expected and ensured a larger than 30 quality score per base. Around 17-26 million RNA-seq reads were generated for each replicate and aligned to the GRCm38/mm10 build of the Mus musculus genome using TopHat v2.0.6 (85). Gene-level expression was quantified as read counts per exon using featureCounts in the Bioconductor package SubRead (86). Reads overlapping exons in annotation build 38.1 of NCBI RefSeq database were included. Counts were converted to log2 counts, fit to a generalized linear model across five wild-type replicates and six mutant samples (data not shown), and normalized to a gene- and sample-specific normalization factor generated by the model with the DESeq2 package (87).

Normalized gene expression values from wild-type E9.5 pSHF RNA-seq data (n = 5) was retrieved for all genes in Figure 5A. For the subcategory of cilia genes, statistical differences in expression level between the DNAHs and IFTs were evaluated through the Mann-Whitney U-test, and for the 26 cilia genes associated with AVSD, the between class statistical differences in expression level according to mutant phenotype were evaluated through the Kruskal-Wallis test followed by Dunn’s multiple comparisons test using GraphPad Prism version 6.05 (GraphPad Software, La Jolla CA, USA).

Statistics

Values are shown as the mean ± SD of the indicated number of measurements. Statistical significance was determined using Student’s t-test (two-tailed) with a significance of 0.05 or FET. No statistical method was used to predetermine sample size, and the animal experiments were not performed in a blinded fashion. Mice were assigned at random to treatment groups for all mouse studies.

Supplementary Material

Supplementary Material is available at HMG online.

Authors’ Contributions

IPM and OBT conceived and designed the study; OBT, JDS, LF, AK, JK and MP performed wet lab experiments; OBT, WH, JS, XHY and IPM participated in data analysis. IPM directed the project; OBT and IPM drafted the manuscript. All authors contributed to the final manuscript.

Acknowledgements

We acknowledge the assistance of Lorenzo Pesce for using the super computer BEAGLE under grant 1S10OD018495-01.

Conflict of Interest statement. None declared.

Funding

This work was supported by National Institutes of Health (R01 HL092153 and R01 HL124836 to I.P.M., T32 GM007183 and HL007381 to J.D.S.) and American Heart Association (13POST17290028 to O.B.T.).

Author notes

†

O.B.T and I.P.M. are co-corresponding authors.